refframe vocabulary

[Markus Demleitner]

Hi,

On Wed, Apr 10, 2019 at 06:14:30PM +0200, Francois Ochsenbein wrote:
> Having recently looked in details in the various coordinate frames, it
> turns out that the definition of the Galactic frame is somewhat fuzzy:
> it was originally defined in the FK4, but Hipparcos and Gaia introduced
> a slightly modified version by fixing the galactic pole and the node
> with angles rounded to 10⁻⁵° directly in the ICRS, and therefore
> differs from the original at the level of ~50mas. And since the
> supergalactic system is defined relative to the galactic, the same
> uncertainty exists for that one. Anyway, the list of the frames will
> likely have to evolve in the near future...

...which of course poses a lot of interesting problems, first among
which is what level of accuracy we want to enable with our annotation
(I'm mentioning E-terms in FK4 and the question of where they were
applied).  50 mas, frankly, sound like quite a challenge to me.

As a sobering reality check, see
http://www.astropy.org/coordinates-benchmark/summary_matrix.html --
that's a comparison of the results of astropy's frame conversion to
those of various other software packages.  I'm quoting "Summary of
differences in arcseconds" from the legend.

Years ago I tried something like this on a much smaller scale, and I
was surprised to see fairly significant disagreements even between
authorities like AST and slalib.

What I'm trying to say: 50 mas through the more exotic
transformations (where I'm surprised that so many libraries agree so
well in fk4->fk5, which is rather non-trivial) and without
significant human intervention seems like a steep plan, and perhaps
we should accompany any refinements in annotation with working out
where the differences between AST and slalib (and pyephem and
astropy) come from and how to fix them.

I guess we should try to get some funds for that...

           -- Markus